Clément Pagés

You need to add a `registry` option containing the path to a database.yml file. For instance: ````bash pyannote-audio-train \ model=PyanNet \ task=VoiceActivityDetection \ registry="AMI-diarization-setup/pyannote/database.yml" \ protocol=AMI.SpeakerDiarization.only_words ```` This should fix...

Could you please share a Google Colab that we can run to reproduce the error (and have more context about when this error occurs)?

> ⚠️ Audio files introduced by this PR are a bit large... All the asset files needed by the tuto have been removed from the repo. Now, these files are...

Hey @jinec , This is a normal behavior. It depends on segments in your `Annotation` object . If there is no segments in `ann`intersecting with `seg`, `spk` will be `None`....

Hey @guijuzhejiang. Did you try to [optimize the pipeline](https://github.com/pyannote/pyannote-audio/blob/develop/tutorials/adapting_pretrained_pipeline.ipynb) on your data ? You can also have a look to the end of this [tuto](https://github.com/pyannote/pyannote-audio/blob/develop/tutorials/overlapped_speech_detection.ipynb). According to your task description,...

Hey @chenfuckthesky, I tested on my side (python 3.8, `pyannote.audio=3.1` and ubuntu 24.04), it's working fine on your test file. Maybe double check the path to your `config.yml`.

Hey @WelkinYang , could you please provide a MRE of your issue (in google colab), so we can reproduce it ?

Hey @eschmidbauer , @RioLLee, could you provide the audios that generated the error so that I can reproduce it? Indeed, I think it's an error that doesn't have the same...

I just tried to reproduce this bug with the following annotations: ``` SPEAKER sample 1 10.570 4.130 speaker90 SPEAKER sample 1 10.570 3.430 speaker91 ``` but I am not able...

Hey @Artpel1805, try to update `pyannote.audio` to 3.3.2, it should solve your issue. (tested with `numpy==1.26.4`)